CN114080903A - Picking type picking execution device and control method thereof - Google Patents

Picking type picking execution device and control method thereof Download PDF

Info

Publication number
CN114080903A
CN114080903A CN202111396456.0A CN202111396456A CN114080903A CN 114080903 A CN114080903 A CN 114080903A CN 202111396456 A CN202111396456 A CN 202111396456A CN 114080903 A CN114080903 A CN 114080903A
Authority
CN
China
Prior art keywords
picking
fixedly connected
arc
camera
article
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202111396456.0A
Other languages
Chinese (zh)
Other versions
CN114080903B (en
Inventor
王知先
邓云蛟
董宇
曾达幸
侯雨雷
王航
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Huicheng Automation Technology Ningbo Co ltd
Original Assignee
Huicheng Automation Technology Ningbo Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Huicheng Automation Technology Ningbo Co ltd filed Critical Huicheng Automation Technology Ningbo Co ltd
Priority to CN202111396456.0A priority Critical patent/CN114080903B/en
Publication of CN114080903A publication Critical patent/CN114080903A/en
Application granted granted Critical
Publication of CN114080903B publication Critical patent/CN114080903B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • AHUMAN NECESSITIES
    • A01AGRICULTURE; FORESTRY; ANIMAL HUSBANDRY; HUNTING; TRAPPING; FISHING
    • A01DHARVESTING; MOWING
    • A01D46/00Picking of fruits, vegetables, hops, or the like; Devices for shaking trees or shrubs
    • A01D46/30Robotic devices for individually picking crops

Landscapes

  • Engineering & Computer Science (AREA)
  • Robotics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Environmental Sciences (AREA)
  • Manipulator (AREA)

Abstract

The invention provides a lifting picking type picking executing device and a control method thereof, wherein the executing device comprises a camera component, a support frame component, a driving mechanism, an intermediate transmission mechanism and a picking executing mechanism, a swing rod is fixedly connected with an output shaft of a steering engine, the working plane of an arc-shaped push block is vertical to the horizontal plane, the integral shape of the arc-shaped push block is arc-shaped, the first end of the arc-shaped push block is fixedly connected with the first end of a slide block connecting rod, a first bearing is in intermittent contact with the working plane of the arc-shaped push block during working, the slide block is in sliding fit with a T-shaped guide rail, a stop block is arranged at the front end of the T-shaped guide rail, two sides of a tension spring are respectively connected with a spring mounting shaft, a walking gear is meshed with a rack, the first end of the rack is fixedly connected with the first end of a main mounting plate, a picking paw is sleeved on a paw stepped shaft and fixedly connected with a walking gear, and the picking paw is made of flexible materials. The tea picking machine can automatically identify objects such as tea tender shoots, can simulate the traditional manual tea picking, cannot cause shearing and extrusion damage to the tea shoots, is exquisite and small in structure, is high in tea picking efficiency, and reduces the manual picking cost.

Description

Picking type picking execution device and control method thereof
Technical Field
The invention relates to the field of agricultural machinery, in particular to a lifting picking type picking execution device and a control method thereof.
Background
Some plants such as famous tea have extremely high requirements on fresh leaves, and the picked fresh leaves are required to have the characteristics of uniformity, tenderness and neatness. Therefore, during the picking process, the fresh leaves need to be picked selectively. At present, most of tea picking is still manual, and in the production of famous tea, because the requirements of fresh leaf raw materials are strict, the labor investment for picking tea in some tea areas accounts for more than 60% of the total production cost, and the tea picking method is the link with the highest labor cost in the tea production process.
And what the mechanized tea plucking that realizes at present generally indicates all is the picking of a large amount of tea, and the tea plucker in present stage is based on shear type principle and picks, and the tea plucker shears tealeaves through the mode of a sword cut fast, though compare artifical the picking and improve picking efficiency, but when this kind of tea plucker picked tealeaves, the phenomenon such as the bud damage, mistake were adopted, was leaked and is adopted very generally, and this kind of problem has influenced further using widely of this kind of tea plucker in tealeaves production field.
With the rapid development of science and technology, robotics has been gradually applied to various industries of human beings; the shortage of rural labor force caused by the aging population and the urbanization construction continuously stimulates the modernization, the unmanned and the intelligent development. Various types of robots are also favored by agricultural production, and are widely used in various fields such as agricultural fertilization, weeding, and picking.
However, in the picking of fresh tea leaves, the number of products of the robot with selective picking is relatively small, so that the development of a picking execution device for the tail end of the robot, which has a delicate structure, can ensure accurate picking and high-speed picking and can simulate manual tea picking, is very necessary.
Disclosure of Invention
In order to solve the defects of the prior art, the invention provides a lifting picking execution device and a control method, which mainly take pictures through a camera component, identify target plants and acquire plane coordinates, acquire depth information of the target plants through a correlation switch, realize lifting picking movement of tea buds by a picking paw through the movement conversion among a driving mechanism, an intermediate transmission mechanism and a picking execution mechanism, and simulate the traditional manual lifting picking type tea picking, so that the high-speed accurate picking of the tea buds is realized, the operation efficiency can be effectively improved, and the labor cost is saved.
The invention provides a lifting picking type picking executing device which comprises a camera component, a support frame component, a driving mechanism, an intermediate transmission mechanism and a picking executing mechanism, wherein the camera component is positioned at the upper part of the support frame component;
the supporting frame assembly comprises an upper mounting flange, an upper connecting frame, a steering engine mounting frame, a lower connecting frame and a bottom main mounting plate, wherein the steering engine mounting frame is arranged between the upper connecting frame and the lower connecting frame;
the driving mechanism comprises a steering engine, a swing rod, a stepped shaft and a first bearing, an output shaft of the steering engine is fixedly connected with the middle of the swing rod, two ends of the swing rod are fixedly connected with a first end of the stepped shaft by means of fasteners, a second end of the stepped shaft is fixedly connected with an inner ring of the first bearing, a limiting platform is arranged near the end part of the second end of the stepped shaft, and the upper surface of the limiting platform is in contact with the lower surface of the first bearing;
the middle transmission mechanism comprises an extension spring and transmission assemblies symmetrically arranged on two sides of the extension spring, each transmission assembly comprises an arc-shaped push block, a slide block connecting rod, a slide block, a connecting block, a spring installation shaft, a T-shaped guide rail and a stop block, the outer end part of each arc-shaped push block is fixedly connected with the first end of each slide block connecting rod, the second end of each slide block connecting rod is fixedly connected with the corresponding slide block, each slide block is in sliding fit with the corresponding T-shaped guide rail, the first end of each connecting block is fixedly connected with the middle position of the inner end part of each arc-shaped push block, the spring installation shaft sequentially penetrates through the arc-shaped push blocks and the connecting blocks and then is connected with the end parts of the extension springs, and the stop blocks are arranged at the front ends of the T-shaped guide rails and are fixedly connected with the first ends of the mounting plates;
picking actuating mechanism includes hand claw step shaft, walking gear, second bearing, rack, flexible picking hand claw and correlation switch, the first end of hand claw step shaft with the second end fixed connection of connecting block, the installation end of hand claw step shaft is provided with the spacing platform of installation, the spacing platform of installation with the connecting block second end contacts, the inner circle of second bearing with the second end fixed connection of hand claw step shaft, the walking gear with the outer lane fixed connection of second bearing just the walking gear with rack toothing, the first end of rack with the first end fixed connection of main mounting panel, flexible picking hand claw with the aid of first connecting piece with the first end fixed connection of walking gear, flexible picking hand claw with the aid of second connecting piece and correlation switch fixed connection.
Preferably, the camera subassembly includes camera mounting flange, C word camera frame and camera, camera mounting flange first end with terminal surface fixed connection under the mounting flange in the support frame subassembly, C word camera frame first end with camera mounting flange second end fixed connection, the camera with C word camera frame second end fixed connection.
Preferably, the two sides of the mounting flange are fixedly connected with the first end of the upper connecting frame respectively, the second end of the upper connecting frame is fixedly connected with the first end of the steering engine mounting frame respectively, the second end of the steering engine mounting frame is fixedly connected with the first end of the lower connecting frame respectively, and the second end of the lower connecting frame is fixedly connected with the first end of the main mounting plate respectively.
Preferably, the working plane of the arc-shaped push block is vertical to the horizontal plane, and the overall shape of the arc-shaped push block is arc-shaped.
Preferably, the first bearings arranged on two sides of the swing rod and the working plane of the arc-shaped push block are located on the same horizontal plane, when the picking device works, the first bearings are in intermittent contact with the working plane of the arc-shaped push block, and the width of the first bearings is smaller than that of the working plane of the arc-shaped push block.
Preferably, the middle rack and the T-shaped guide rail are arranged in parallel, the length of the T-shaped guide rail is equal to that of the rack, and the length of the T-shaped guide rail is greater than a half of a difference between a maximum width of the first bearings mounted on two sides of the swing rod and a distance between the blocking pieces.
Preferably, the upper connecting frame and the steering engine mounting frame are symmetrically distributed relative to the steering engine, the lower connecting frame is symmetrically distributed on two sides of the device relative to the main mounting plate, and the middle transmission mechanism and the picking actuating mechanism are symmetrically distributed on two sides of the device.
Preferably, the first connecting piece is a flange, the second connecting piece is a vertical plate, the vertical plate is fixedly connected with the correlation switch, and the correlation switch is provided with correlation signal holes for the correlation switches at two sides to transmit and receive signals.
Preferably, the invention also discloses a control method of the lifting picking execution device, which comprises the following steps:
s1, horizontally positioning the article: acquiring horizontal coordinates of an area where the object is located based on a real-time picture shot by a camera, further controlling a picking execution device to move to the position above the area to be picked, and judging the depth information of the target object by a correlation switch;
s2, identifying the type of the article, which comprises the following substeps:
s21, building an article type database according to the type of the article: marking the photos of the articles collected in advance to obtain an article type database, and marking different articles as different pixel values;
s22, training of the semantic segmentation network based on the UNet image: segmenting the network based on the UNet image semantic meaning of the U-shaped network and training the article type database obtained in the step S21 to obtain a high-precision type recognition algorithm, which specifically comprises the following steps:
s221, splitting the article variety database according to the ratio of the training set to the test set of 9:1 to obtain a training set and a test set which comprise an original image and a label image;
s222, selecting an adaptive matrix analysis Adam optimizer, and performing iterative training based on a UNet image semantic segmentation network comprising an encoder and a decoder;
s223, calculating a confusion matrix H between a training result graph of each original image in the training set and the corresponding label graph in the test set:
Figure BDA0003370496020000041
wherein TP represents the number of pixel points of various pixel values of the article in the training result, which are consistent with the corresponding pixel values in the label graph; FP represents the number of pixel points of which the background pixel values in the training result are inconsistent with the corresponding pixel values in the label graph; FN represents the number of pixel points of each type of pixel value of the article in the training result, which is inconsistent with the corresponding pixel value in the label graph; TN represents the pixel point number of the background pixel value in the training result consistent with the corresponding pixel value in the label graph;
s224, calculating the intersection ratio IoU of each training result graph in the training set and the test set based on the confusion matrix H, wherein the calculation formula is as follows:
Figure BDA0003370496020000042
s225, calculating the average MIoU of the intersection ratio IoU of all training result graphs in the training set and the test set, judging the average MIoU of IoU, if the average MIoU of IoU is larger than or equal to a judgment threshold, directly outputting a high-precision type recognition algorithm weight file to obtain a high-precision type recognition algorithm, and if the average MIoU of IoU is smaller than the judgment threshold, repeatedly executing the steps S222-S224 until the average MIoU of IoU reaches the judgment threshold, outputting the high-precision type recognition algorithm weight file to obtain the high-precision type recognition algorithm;
s3, identifying the type of the article: identifying and classifying the articles based on the high-precision type identification algorithm obtained in the step S2 and by combining the real-time pictures shot by the camera in the step S1, the method specifically comprises the following steps:
s31, inputting the real-time picture shot by the camera in the step S1 and the high-precision type recognition algorithm weight file obtained in the step S2 into a UNet image semantic segmentation network to obtain a segmentation result graph;
s32, determining the type of the article based on the pixel value of the area where the article is located in the segmentation result graph;
and S4, after the type of the article is determined, whether the determined type of the article belongs to the picking target is determined, if so, the picking execution device is controlled to pick.
Preferably, the determination threshold in step S225 is 95%.
Compared with the prior art, the invention has the following beneficial effects:
(1) according to the invention, the intermittent contact type cam double-push-rod mechanism is adopted, the swing rod and the first bearings arranged on two sides form the rod-shaped cam, and the arc-shaped push blocks on two sides are the push blocks.
(2) According to the invention, through the arrangement of the intermediate transmission mechanism and the picking actuating mechanism, the rotary motion provided by the single steering engine is converted into the compound motion of translation and rotation of the two flexible picking claws, and the two flexible picking claws are matched to realize clamping and picking actions, so that the traditional manual tea picking is simulated, the picking claws select flexible materials according to the static friction force required by picking and the mechanical properties of tensile strength and compressive strength of picked articles, the cutting and extrusion damage to tea buds can not be caused, and the completeness and quality of the picked articles can be ensured.
(3) The automatic tea bud picking machine can automatically identify and position objects such as target tea buds, is exquisite and exquisite in structure, can adapt to picking of tea buds of single tea buds, one tea bud and two tea buds of each tea bud, can realize quick and accurate picking of the target tea buds by matching with the multi-degree-of-freedom mechanical arm, can realize long-distance collection and temporary storage by matching with the negative pressure suction device, improves tea picking efficiency, relieves the problem of delayed tea bud picking caused by insufficient labor force, and reduces manual picking cost.
Drawings
FIG. 1 is a schematic view of the overall structure of the present invention;
FIG. 2 is a schematic view of a camera assembly according to the present invention;
FIG. 3 is a schematic view of a support bracket assembly of the present invention;
FIG. 4 is an exploded view of the drive mechanism of the present invention;
FIG. 5 is a schematic view of an intermediate transmission mechanism of the present invention;
fig. 6 is an exploded view of the picking actuator of the present invention.
The main reference numbers are as follows:
the device comprises a camera component 1, a camera mounting flange 11, a C-shaped camera frame 12, a camera 13, a support frame component 2, a mounting flange 21, an upper connecting frame 22, a steering engine mounting frame 23, a lower connecting frame 24, a main mounting plate 25, a driving mechanism 3, a steering engine 31, a swing rod 32, a stepped shaft 33, a first bearing 34, an intermediate transmission mechanism 4, an arc-shaped push block 41, a slider connecting rod 42, a slider 43, a connecting block 44, a spring connecting shaft 45, a tension spring 46, a T-shaped guide rail 47, a stop block 48, a picking executing mechanism 5, a paw stepped shaft 51, a walking gear 52, a second bearing 53, a rack 54, a flexible picking paw 55, a paw flange 551, a vertical plate 552, an opposite emission signal hole 553 and an opposite emission switch 56.
Detailed Description
Hereinafter, embodiments of the present invention will be described with reference to the drawings.
The device of the invention is suitable for picking in a lifting mode, and is particularly suitable for picking of tender tea leaves.
The lifting picking type picking execution device and the control method thereof provided by the invention are shown in fig. 1 and comprise a camera component 1, a support frame component 2, a driving mechanism 3, an intermediate transmission mechanism 4 and a picking execution mechanism 5. Specifically, the camera assembly 1 is located on the upper portion of the support frame assembly 2, the driving mechanism 3 is connected with the middle portion of the support frame assembly 2, the driving mechanism 3 is connected with the upper portion of the support frame assembly 2, the support frame assembly 2 supports the internal structure of the whole device, the middle transmission mechanisms 4 are located on two sides of the lower portion of the driving mechanism 3, and the picking execution mechanism 5 is connected with the middle transmission mechanisms 4.
As shown in fig. 2, the camera assembly 1 includes a camera mounting flange 11, a C-shaped camera frame 12 and a camera 13, a first end of the camera mounting flange 11 is fixedly connected with a lower end face of a mounting flange 21 in the support frame assembly, a first end of the C-shaped camera frame 12 is fixedly connected with a second end of the camera mounting flange 11, the camera 13 is fixedly connected with a second end of the C-shaped camera frame 12, and the camera 13 is responsible for image acquisition of a tea tent surface in an area, so as to obtain an image of an object such as the tea tent surface, wherein the image of the tea tent surface includes a plurality of tea buds to be picked.
As shown in fig. 3, the support frame assembly 2 includes a mounting flange 21, an upper connecting frame 22, a steering engine mounting frame 23, a lower connecting frame 24 and a main mounting plate 25, wherein mounting holes are formed on the end surface of the mounting flange 21 so as to be connected with the tail end of the robot, for other robots of different models, different flanges or mounting plates of other shapes can be designed to be matched with the tail end of the robot, the two sides of the mounting flange 21 are respectively and fixedly connected with the first end of the upper connecting frame 22, the second end of the upper connecting frame 22 is respectively and fixedly connected with the first end of the steering engine mounting frame 23, the second end of the steering engine mounting frame 23 is respectively and fixedly connected with the first end of the lower connecting frame 24, the first end of the main mounting plate 25 is respectively and fixedly connected with the second end of the lower connecting frame 24, wherein the upper connecting frame 22 and the steering engine mounting frame 23 are symmetrically distributed relative to the steering engine 31, the lower connecting frame 24 in the support frame assembly is symmetrically distributed relative to the two sides of the main mounting plate 25, and the number is set to two.
As shown in fig. 4, the driving mechanism 3 includes a steering engine 31, a swing rod 32, a stepped shaft 33 and a first bearing 34, a power input source of the picking device may be the steering engine 31 or other motors, an output shaft of the steering engine 31 is fixedly connected with the middle part of the swing rod 32 through a fastening bolt, two ends of the swing rod 32 are respectively fixedly connected with a first end of the stepped shaft 33 through fastening holes, a second end of the stepped shaft 33 is fixedly connected with an inner ring of the first bearing 34, a limiting platform is arranged at a lower end of the stepped shaft 33, and an upper surface of the limiting platform contacts with a lower surface of the first bearing 34.
As shown in fig. 5, the intermediate transmission mechanism 4 includes an extension spring 46 and transmission components disposed at two sides of the extension spring 46, the transmission components are symmetrically disposed and include an arc-shaped push block 41, a slider link 42, a slider 43, a connection block 44, a spring mounting shaft 45, a T-shaped guide rail 47 and a stop block 48, wherein the working plane of the arc-shaped push block 41 is perpendicular to the horizontal plane, and the overall shape is arc-shaped, the first end of the arc-shaped push block is fixedly connected with the first end of the slider link 42 through a mounting hole, the slider 43 is fixedly connected with the second end of the slider link 42, the slider 43 is in sliding fit with the T-shaped guide rail 47, the length of the T-shaped guide rail 47 and the length of the rack 54 in the picking execution mechanism 5 are slightly greater than the difference between the maximum width between the first bearings 34 mounted at two sides of the swing link 32 and the distance between the stop blocks 48, the first end of the connection block 44 is fixedly connected with the middle of the second end of the arc-shaped push block 41, the spring mounting shaft 45 passes through the arc-shaped push block 41 and the connection block 44 and is preset and fixedly connected with both, the two sides of the extension spring 46 are respectively connected with the spring mounting shaft 45, the extension spring 46 is used for retracting the two arc-shaped push blocks 41 pushed away by the swing rod 32, the stop 48 is arranged at the front end of the T-shaped guide rail 47 and fixedly connected with the first end of the mounting plate 25, and the stop 48 is used for limiting the initial positions of the two arc-shaped push blocks 41.
Further, the working planes of the first bearings 34 and the arc-shaped push blocks 41 arranged on the two sides of the swing rod 32 are located on the same horizontal plane, when the picking device works, the working planes of the first bearings 34 and the arc-shaped push blocks 41 are in intermittent contact, and the width of the first bearings 34 is slightly smaller than that of the working planes of the arc-shaped push blocks 41. The initial positions of the two arc-shaped push blocks 41 are at a certain distance, when the swing rod 32 drives the first bearings 34 at the two sides to do the whole-circle motion, the two arc-shaped push blocks 41 are touched and pushed intermittently, so that the two arc-shaped push blocks 41 move towards the two sides of the picking device along the T-shaped guide rail 47, when the swing rod 32 is parallel to the T-shaped guide rail 47, the arc-shaped push blocks 41 move to the farthest distance, at the moment, the swing rod 32 continues to rotate, and the two arc-shaped push blocks 41 also return to the stop block 48 under the action of the extension spring 46.
As shown in fig. 6, the picking actuator 5 comprises a claw stepped shaft 51, a walking gear 52, a second bearing 53, a rack 54, a flexible picking claw 55 and an opposite switch 56, wherein a first end of the claw stepped shaft 51 is fixedly connected with a second end of the connecting block 44, a mounting end of the claw stepped shaft 51 is provided with a mounting limit platform and contacts with the second end of the connecting block 44, an inner ring of the second bearing 53 is fixedly connected with the second end of the claw stepped shaft 51, the walking gear 52 is fixedly connected with an outer ring of the second bearing 53, the walking gear 52 is meshed with the rack 54, a first end of the rack 54 is fixedly connected with a first end of the main mounting plate 25, the flexible picking claw 55 is provided with a flange 551, the flexible picking claw 55 is freely sleeved on the claw stepped shaft 51 and is fixedly connected with the first end of the walking gear 52 through the flange, the flexible picking claw 55 is provided with a vertical plate 552, the opposite switch 56 is fixedly connected with a mounting hole of the vertical plate 552, picking paw 55 is equipped with correlation signal hole 553 and supplies both sides correlation switch 56 to transmit the received signal, and correlation switch 56 is used for controlling means's start-up, and flexible picking paw 55 carries out the selection of flexible material according to the static friction force that picks required and the tensile, the resistance to compression mechanical properties of picking article, when picking article such as tealeaves, is used for guaranteeing that the tea bud of picking can not receive the extrusion damage to ensure the goodness of the tea bud of picking.
Further, the T-shaped guide rail 47 in the intermediate transmission mechanism 4 and the rack 54 in the picking actuator 5 are parallel to each other, the length of the T-shaped guide rail 47 is equal to the length of the rack 54, the length of the T-shaped guide rail is slightly greater than half of the difference between the maximum width between the first bearings 34 installed on the two sides of the swing rod 32 and the distance between the stoppers 48, and the intermediate transmission mechanism 4 and the picking actuator 5 are symmetrically distributed on the two sides of the device, and the number of the intermediate transmission mechanism 4 and the picking actuator 5 is two.
The control method of the picking device takes famous tea as an example, the control method of the famous tea picking execution device comprises a famous tea horizontal positioning algorithm and a famous tea type identification algorithm, the famous tea horizontal positioning algorithm obtains the horizontal coordinate of the region where famous tea is located based on the real-time picture shot by the camera 13, then the picking execution device is controlled to move to the position above the region to be picked, the depth information of the target tea bud is judged by the correlation switch 56, when the tea bud enters the middle of the two flexible picking claws 55, the signal emitted by the emitting end of the correlation switch 56 can be blocked, and the correlation switch 56 sends a starting signal to the steering engine 31.
Specifically, the famous tea variety recognition algorithm comprises the steps of building a famous tea variety database, training based on a UNet image semantic segmentation network and recognizing the famous tea variety; an implementation of identifying an item category, comprising the sub-steps of: according to the type of the article, an article type database is built: marking pre-collected photos of articles to obtain an article type database, wherein different articles are marked as different pixel values; training of a semantic segmentation network based on UNet images: based on a UNet image semantic segmentation network of a U-shaped network, training by combining an article type database to obtain a high-precision type recognition algorithm; splitting the article category database according to a training and testing ratio of 9:1 to obtain a training set and a testing set which comprise an original image and a label image; an Adam optimizer for adaptive moment analysis is selected, and iterative training is carried out on the basis of a UNet image semantic segmentation network comprising an encoder and a decoder; calculating a confusion matrix H between a training result graph of each original image in the training set and the test set and a corresponding label graph:
Figure BDA0003370496020000091
wherein TP represents the number of pixel points of various pixel values of the article in the training result, which are consistent with the corresponding pixel values in the label graph; FP represents the number of pixel points of which the background pixel values in the training result are inconsistent with the corresponding pixel values in the label graph; FN represents the number of pixel points of each type of pixel value of the article in the training result, which is inconsistent with the corresponding pixel value in the label graph; TN represents the pixel point number of the background pixel value in the training result consistent with the corresponding pixel value in the label graph;
calculating the intersection ratio IoU of each training result graph in the training set and the test set based on the confusion matrix H:
Figure BDA0003370496020000092
calculating the mean value MIoU of the intersection ratio IoU of all the training result graphs in the training set and the test set, and repeatedly executing the step S222 until the mean value MIoU reaches more than 95%, outputting a weight file of the high-precision type recognition algorithm, namely obtaining the high-precision type recognition algorithm; the identification of the type of article can then be carried out: identifying and classifying the articles based on a high-precision type identification algorithm and in combination with the real-time pictures taken by the camera 13 in the step S1; inputting the real-time picture shot by the camera in the step S1 and the high-precision type recognition algorithm weight file acquired in the step S2 into a UNet image semantic segmentation network to acquire a segmentation result graph; and determining the type of the article based on the pixel value of the area where the article is located in the segmentation result graph.
The following describes a picking-type picking actuator and a control method thereof in combination with an embodiment:
as the famous tea tender shoots are mostly kept to grow vertically under the action of sunlight, the flexible picking paw 55 only needs to be kept in a state vertical to the ground when picking, namely, the best picking pose is obtained, and therefore, as the tail end execution device, the picking device does not need to have other degrees of freedom besides the lifting picking action.
The camera component 1, the support frame component 2, the driving mechanism 3, the intermediate transmission mechanism 4 and the picking actuating mechanism 5 are firstly installed according to the connection relation of the structures, then the components are sequentially installed on the support frame component 2 according to the sequence of the driving mechanism 3, the intermediate transmission mechanism 4 and the picking actuating mechanism 5, and finally the camera component 1 is installed. After the assembly of the picking device is completed, the picking device is installed at the tail end of the serial or parallel multi-shaft mechanical arm on the gantry type crawler trolley fixed platform through the installation flange 21, the picking device is guaranteed to move flexibly and can reach all-directionally in the installation process, high-speed non-damage picking of tea tender shoots is achieved, and picking efficiency and quality are guaranteed.
Then, the whole picking device is driven to move to a position to be picked on the tea tent by the multi-shaft mechanical arm on the gantry type crawler trolley, then the camera 13 acquires tea image data of one area, a target tea bud is identified based on a UNet image semantic segmentation network algorithm, a two-dimensional plane coordinate of the target tea bud is acquired by a horizontal positioning algorithm, the multi-shaft mechanical arm changes the position and posture according to the acquired target coordinate, so that the two flexible picking claws 55 on the picking device move to the position above the target tea bud, the mechanical arm drives the picking device to vertically descend, and depth information is acquired by the correlation switch 56.
When a target tea bud enters between the two flexible picking claws 55, due to the blockage of the tea bud, a receiving end of the opposite-jet switch 56 cannot receive a signal transmitted from a transmitting end, at the moment, the opposite-jet switch 56 transmits a starting signal to the steering engine 31, the steering engine 31 drives the swing rod 32 to rotate, the first bearing 34 arranged at the tail end of the swing rod 32 contacts the arc-shaped push block 41 after the swing rod 32 rotates for a certain angle, the arc-shaped push block 41 pushes the slider connecting rod 42 to move towards two sides along the T-shaped guide rail 47 under the action of the first bearing 34, at the moment, the walking gear 52 is meshed with the rack 54 to realize simultaneous rotating motion while moving towards two sides along with the claw stepped shaft 51, the two flexible picking claws 55 are matched to realize picking motion, when the swing rod 32 is parallel to the T-shaped guide rail 47, the arc-shaped push block 41 moves to the farthest distance, at the moment, the steering engine 31 drives the swing rod 32 to continue rotating, the two arc-shaped push blocks 41 are also reset to the stop block 48 under the action of the extension spring 46, the flexible picking paw 55 is restored to the initial position, one-time picking action is completed, the picked tea buds can be collected and stored in a longer distance by matching with the negative pressure suction device, and the collected tea buds are moved to the next picking target in the area under the drive of the multi-degree-of-freedom mechanical arm to perform repeated operation.
The above-mentioned embodiments are merely illustrative of the preferred embodiments of the present invention, and do not limit the scope of the present invention, and various modifications and improvements made to the technical solution of the present invention by those skilled in the art without departing from the spirit of the present invention shall fall within the protection scope defined by the claims of the present invention.

Claims (10)

1. The utility model provides an actuator is picked to extraction formula which characterized in that: the picking device comprises a camera component, a support frame component, a driving mechanism, a middle transmission mechanism and a picking execution mechanism, wherein the camera component is positioned at the upper part of the support frame component, the driving mechanism is connected with the middle part of the support frame component, the middle transmission mechanism is positioned at two sides of the lower part of the driving mechanism, and the picking execution mechanism is connected with the middle transmission mechanism;
the supporting frame assembly comprises an upper mounting flange, an upper connecting frame, a steering engine mounting frame, a lower connecting frame and a bottom main mounting plate, wherein the steering engine mounting frame is arranged between the upper connecting frame and the lower connecting frame;
the driving mechanism comprises a steering engine, a swing rod, a stepped shaft and a first bearing, an output shaft of the steering engine is fixedly connected with the middle of the swing rod, two ends of the swing rod are fixedly connected with a first end of the stepped shaft by means of fasteners, a second end of the stepped shaft is fixedly connected with an inner ring of the first bearing, a limiting platform is arranged near the end part of the second end of the stepped shaft, and the upper surface of the limiting platform is in contact with the lower surface of the first bearing;
the middle transmission mechanism comprises an extension spring and transmission assemblies symmetrically arranged on two sides of the extension spring, each transmission assembly comprises an arc-shaped push block, a slide block connecting rod, a slide block, a connecting block, a spring installation shaft, a T-shaped guide rail and a stop block, the outer end part of each arc-shaped push block is fixedly connected with the first end of each slide block connecting rod, the second end of each slide block connecting rod is fixedly connected with the corresponding slide block, each slide block is in sliding fit with the corresponding T-shaped guide rail, the first end of each connecting block is fixedly connected with the middle position of the inner end part of each arc-shaped push block, the spring installation shaft sequentially penetrates through the arc-shaped push blocks and the connecting blocks and then is connected with the end parts of the extension springs, and the stop blocks are arranged at the front ends of the T-shaped guide rails and are fixedly connected with the first ends of the mounting plates;
picking actuating mechanism includes hand claw step shaft, walking gear, second bearing, rack, flexible picking hand claw and correlation switch, the first end of hand claw step shaft with the second end fixed connection of connecting block, the installation end of hand claw step shaft is provided with the spacing platform of installation, the spacing platform of installation with the connecting block second end contacts, the inner circle of second bearing with the second end fixed connection of hand claw step shaft, the walking gear with the outer lane fixed connection of second bearing just the walking gear with rack toothing, the first end of rack with the first end fixed connection of main mounting panel, flexible picking hand claw with the aid of first connecting piece with the first end fixed connection of walking gear, flexible picking hand claw with the aid of second connecting piece and correlation switch fixed connection.
2. The extraction picking actuator of claim 1, wherein: the camera subassembly includes camera mounting flange, C word camera frame and camera, camera mounting flange first end with terminal surface fixed connection under the mounting flange in the support subassembly, C word camera frame first end with camera mounting flange second end fixed connection, the camera with C word camera frame second end fixed connection.
3. The extraction picking actuator of claim 1, wherein: the mounting flange is characterized in that the two sides of the mounting flange are fixedly connected with the first end of the upper connecting frame respectively, the second end of the upper connecting frame is fixedly connected with the first end of the steering engine mounting frame respectively, the second end of the steering engine mounting frame is fixedly connected with the first end of the lower connecting frame respectively, and the second end of the lower connecting frame is fixedly connected with the first end of the main mounting plate respectively.
4. The extraction picking actuator of claim 1, wherein: the working plane of the arc-shaped push block is vertical to the horizontal plane, and the overall shape of the arc-shaped push block is arc-shaped.
5. The extraction picking actuator of claim 1, wherein: the first bearings arranged on two sides of the swing rod and the working plane of the arc-shaped push block are located on the same horizontal plane, when the picking device works, the first bearings are in intermittent contact with the working plane of the arc-shaped push block, and the width of the first bearings is smaller than that of the working plane of the arc-shaped push block.
6. The extraction picking actuator of claim 1, wherein: the rack and the T-shaped guide rail are arranged in parallel, the length of the T-shaped guide rail is equal to that of the rack, and the length of the T-shaped guide rail is greater than half of the difference between the maximum width of the first bearings arranged on two sides of the swing rod and the distance between the stop blocks.
7. The extraction picking actuator of claim 1, wherein: the upper connecting frame and the steering engine mounting frame are symmetrically distributed relative to the steering engine, the lower connecting frame is symmetrically distributed on two sides of the device relative to the main mounting plate, and the middle transmission mechanism and the picking executing mechanism are symmetrically distributed on two sides of the device.
8. The extraction picking actuator of claim 1, wherein: the first connecting piece is a flange plate, the second connecting piece is a vertical plate, the vertical plate is fixedly connected with the correlation switch, and the correlation switch is provided with correlation signal holes for the correlation switches at two sides to transmit and receive signals.
9. A control method of a lifting picking execution device is characterized in that: which comprises the following steps:
s1, horizontally positioning the article: acquiring horizontal coordinates of an area where the object is located based on a real-time picture shot by a camera, further controlling a picking execution device to move to the position above the area to be picked, and judging the depth information of the target object by a correlation switch;
s2, identifying the type of the article, which comprises the following substeps:
s21, building an article type database according to the type of the article: marking pre-collected photos of articles to obtain an article type database, and marking different articles as different pixel values;
s22, training of the semantic segmentation network based on the UNet image: segmenting the network based on the UNet image semantic meaning of the U-shaped network and training the article type database obtained in the step S21 to obtain a high-precision type recognition algorithm, which specifically comprises the following steps:
s221, splitting the article variety database according to the ratio of the training set to the test set of 9:1 to obtain a training set and a test set which comprise an original image and a label image;
s222, selecting an adaptive matrix analysis Adam optimizer, and performing iterative training based on a UNet image semantic segmentation network comprising an encoder and a decoder;
s223, calculating a confusion matrix H between a training result graph of each original image in the training set and the corresponding label graph in the test set:
Figure FDA0003370496010000031
wherein TP represents the number of pixel points of various pixel values of the article in the training result, which are consistent with the corresponding pixel values in the label graph; FP represents the number of pixel points of which the background pixel values in the training result are inconsistent with the corresponding pixel values in the label graph; FN represents the number of pixel points of each type of pixel value of the article in the training result, which is inconsistent with the corresponding pixel value in the label graph; TN represents the pixel point number of the background pixel value in the training result consistent with the corresponding pixel value in the label graph;
s224, calculating the intersection ratio IoU of each training result graph in the training set and the test set based on the confusion matrix H, wherein the calculation formula is as follows:
Figure FDA0003370496010000032
s225, calculating the average MIoU of the intersection ratio IoU of all training result graphs in the training set and the test set, judging the average MIoU of IoU, if the average MIoU of IoU is larger than or equal to a judgment threshold, directly outputting a high-precision type recognition algorithm weight file to obtain a high-precision type recognition algorithm, and if the average MIoU of IoU is smaller than the judgment threshold, repeatedly executing the steps S222-S224 until the average MIoU of IoU reaches the judgment threshold, outputting the high-precision type recognition algorithm weight file to obtain the high-precision type recognition algorithm;
s3, identifying the type of the article: identifying and classifying the articles based on the high-precision type identification algorithm obtained in the step S2 and by combining the real-time pictures shot by the camera in the step S1, the method specifically comprises the following steps:
s31, inputting the real-time picture shot by the camera in the step S1 and the high-precision type recognition algorithm weight file obtained in the step S2 into a UNet image semantic segmentation network to obtain a segmentation result graph;
s32, determining the type of the article based on the pixel value of the area where the article is located in the segmentation result graph;
and S4, after the article type is determined, whether the determined article type belongs to the picking target is determined, if so, the picking execution device is controlled and controlled to pick.
10. The control method of an extraction picking actuator according to claim 9, characterised in that: in step S225, the threshold is determined to be 95%.
CN202111396456.0A 2021-11-23 2021-11-23 Lifting picking type picking execution device and control method thereof Active CN114080903B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111396456.0A CN114080903B (en) 2021-11-23 2021-11-23 Lifting picking type picking execution device and control method thereof

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111396456.0A CN114080903B (en) 2021-11-23 2021-11-23 Lifting picking type picking execution device and control method thereof

Publications (2)

Publication Number Publication Date
CN114080903A true CN114080903A (en) 2022-02-25
CN114080903B CN114080903B (en) 2022-08-19

Family

ID=80303384

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111396456.0A Active CN114080903B (en) 2021-11-23 2021-11-23 Lifting picking type picking execution device and control method thereof

Country Status (1)

Country Link
CN (1) CN114080903B (en)

Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102428802A (en) * 2011-09-14 2012-05-02 西北农林科技大学 Kiwi picking end effector
JP2013240843A (en) * 2012-05-18 2013-12-05 Toyota Industries Corp Finger mechanism for robot hand, and robot hand provided with the finger mechanism
CN206263981U (en) * 2016-12-14 2017-06-20 上海通彩机器人有限公司 A kind of cylinder mechanical grip
CN106863342A (en) * 2017-03-23 2017-06-20 南京工程学院 A kind of terminal executor of picking robot
CN207104915U (en) * 2017-07-18 2018-03-16 金宇芳 A kind of spring clamp device of spring reclaimer robot
CN109500834A (en) * 2018-11-28 2019-03-22 东莞理工学院 A kind of end effector mechanism of picking mechanical arm
CN109845490A (en) * 2019-01-29 2019-06-07 重庆文理学院 Suitable for clamping the end effector of spherical fruit
CN109952859A (en) * 2019-04-17 2019-07-02 北京石油化工学院 Screw-on pineapple picker
CN110640771A (en) * 2019-09-11 2020-01-03 哈尔滨工程大学 Mechanical claw for grabbing bar underwater
CN210589338U (en) * 2019-09-29 2020-05-22 河北农业大学 Tomato picking robot multicavity formula gasbag end effector
CN214691988U (en) * 2021-01-22 2021-11-12 惠怀林 Clamping jaw type taking and unloading device with rotation and revolution

Patent Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102428802A (en) * 2011-09-14 2012-05-02 西北农林科技大学 Kiwi picking end effector
JP2013240843A (en) * 2012-05-18 2013-12-05 Toyota Industries Corp Finger mechanism for robot hand, and robot hand provided with the finger mechanism
CN206263981U (en) * 2016-12-14 2017-06-20 上海通彩机器人有限公司 A kind of cylinder mechanical grip
CN106863342A (en) * 2017-03-23 2017-06-20 南京工程学院 A kind of terminal executor of picking robot
CN207104915U (en) * 2017-07-18 2018-03-16 金宇芳 A kind of spring clamp device of spring reclaimer robot
CN109500834A (en) * 2018-11-28 2019-03-22 东莞理工学院 A kind of end effector mechanism of picking mechanical arm
CN109845490A (en) * 2019-01-29 2019-06-07 重庆文理学院 Suitable for clamping the end effector of spherical fruit
CN109952859A (en) * 2019-04-17 2019-07-02 北京石油化工学院 Screw-on pineapple picker
CN110640771A (en) * 2019-09-11 2020-01-03 哈尔滨工程大学 Mechanical claw for grabbing bar underwater
CN210589338U (en) * 2019-09-29 2020-05-22 河北农业大学 Tomato picking robot multicavity formula gasbag end effector
CN214691988U (en) * 2021-01-22 2021-11-12 惠怀林 Clamping jaw type taking and unloading device with rotation and revolution

Also Published As

Publication number Publication date
CN114080903B (en) 2022-08-19

Similar Documents

Publication Publication Date Title
Williams et al. Robotic kiwifruit harvesting using machine vision, convolutional neural networks, and robotic arms
Hayashi et al. Robotic harvesting system for eggplants
Foglia et al. Agricultural robot for radicchio harvesting
Van Henten et al. An autonomous robot for de-leafing cucumber plants grown in a high-wire cultivation system
CN107696047B (en) End effector of cucumber flexible picking manipulator and picking method
CN107309876B (en) Control method for picking by mechanical arm
CN111418349B (en) Intelligent fruit picking robot and method for realizing fruit picking
CN109328973A (en) A kind of intelligent system and its control method of tapping rubber of rubber tree
CN108406887A (en) A kind of strip garden stuff processing device and processing method based on image vision technology
CN109380146B (en) Automatic measurement device and method for live pigs
CN110916221B (en) Automatic orange ball sectioning device and method
CN111684924A (en) Strawberry picking robot based on raspberry group
CN114080903B (en) Lifting picking type picking execution device and control method thereof
Anh et al. Developing robotic system for harvesting pineapples
CN115194742A (en) Non-contact type automatic tomato stringing picking manipulator and picking method
CN108370740B (en) Topping end effector and topping machine comprising same
CN114830915A (en) Litchi vision picking robot based on laser radar navigation and implementation method thereof
CN104970082A (en) Automatic freshwater mussel opening device
CN211020054U (en) Fruit and vegetable picking device
CN207219437U (en) A kind of strawberry picking vehicle
Hu et al. Design and experiment of a new citrus harvesting robot
CN116897717A (en) Tomato threshing robot end effector and threshing method
CN217184250U (en) Full-automatic high flux group of potato tissue culture seedling banks up rapid propagation device with earth
CN114568306A (en) Full-automatic high flux group of potato tissue culture seedling banks up rapid propagation device with earth
CN112329697B (en) Improved YOLOv 3-based on-tree fruit identification method

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant